Does Okun’s Law Exist in Nigeria? Evidence from the ARDL Bounds Testing Approach

This study employs the autoregressive distributed lag (ARDL) bounds testing technique to examine whether Okun’s law exists in Nigeria during 1970-2014. In addition, this study considers the role of oil prices in the Nigerian economy. The empirical results indicate that a cointegrating or long term relationship exists between the unemployment rate, economic growth and oil prices. In addition, the results demonstrate that in Nigeria, in the long term, unemployment has a negative and significant effect on economic growth, and oil prices have a significant and positive effect on economic growth. The coefficient of unemployment (0.18%) for this study is far less than the result reported by Okun and other studies that focused on developed countries. This suggests that the Okun coefficient is not only unstable but varies for different countries, and does not remain constant for Nigeria. However, policymakers should take steps to reduce unemployment to enhance economic growth in Nigeria. O04,

Interestingly, unimpressive growth rates and high unemployment rates in Nigeria have attracted scholars to investigate the relationship between these two economic indicators (see Abraham, 2014;Akeju & Olanipekun, 2015;Bankole & Fatai, 2013;Jibir, Bappayaya, & Babayo, 2015). These studies are beneficial, however, they have certain shortcomings. For example, Bankole and Fatai (2013) and Abraham (2014) used 29 years of annual data in their analysis. This falls short of the requirement for a time series analy-  1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998  were not conducted, and only one stability test was conducted when, generally, two tests are required for an ARDL analysis. Third, a major shortcoming of the study conducted by Jibir et al. (2015) is that the estimates generated using the ordinary least squares (OLS) estimator, may be biased and inefficient because the unit root property of the variables was not tested and post estimation tests such as serial correlation or heteroscedasticity were not conducted.
Furthermore, the Johansen cointegration method employed by Akeju and Olanipekun (2015) requires a large sample size for validity because it has low power for small samples (Ozturk & Acaravci, 2010 Jibir et al. (2015), who reported that the unemployment rate has a negative and significant effect on economic growth in Nigeria. The coefficient for this study is less than Okun's coefficient and coefficients that have been reported by other studies that focused on developed countries.
The remainder of this paper is organized as follows.
Section two includes a review of empirical studies regarding Okun's law and section three includes the theoretical framework. Section four describes the method and data, and section five provides the results and discussion. Section six concludes the paper. Okun's law has been investigated at country-specific levels, particularly for developed countries. For example, Evans (1989) determined that a feedback relationship exists between the stochastic components of economic growth and the unemployment rate for the United States during 1950-1985. Farsio and Quade (2003 empirically investigated the connection between unemployment and GDP using twenty years of quarterly data for the United States. Using simple regression, the authors determined that the unemployment rate has a significant and negative effect on economic growth, however, the coefficient estimate is much less than Okun's estimate. Furthermore, the unemployment rate Granger causes economic growth, and this relationship was stable for various sub-periods. The authors concluded that Okun's coefficient varies for countries and across different samples. Hamada and Kurosaka (1984)  Additionally, Okun's coefficients quantitatively differ across the regions. The results suggest that disparities may be partially explained by regional differences in productivity growth. Huang and Yeh (2013)  Notably the few studies regarding the unemployment-output nexus in Nigeria used inadequate sample data (see Abraham, 2014;Bankole & Fatai, 2013), employed inappropriate estimation methods or failed to conduct important tests (see Jibir et al., 2015). Jibir et al. (2015) should not have used the OLS method because some of the data used for analysis had a unit root. Furthermore, the authors did not conduct post estimation tests such as serial correlation and heteroscedasticity tests to ensure that the estimates were free from these problems. In addition, Osinubi (2005) did not perform a unit root test on the data used in his study. Therefore, this study fills a gap by estimating the Okun-type relationship in Nigeria using a recently developed ARDL bounds testing technique and conducting all the important tests (unit root test, diagnostic tests, etc.,) in addition to expanding the scope of the study (by analyzing, the data for 1970-2014).

Theoretical framework
This section establishes a connection between unemployment and output. Researchers have relied on Okun's law when examining the link between the unemployment rate and output growth. Okun's law can be explained as follows. As aggregate demand changes, firms/producers change their output plans, which causes variations in the demand for labor and subsequently, changes unemployment rates.
According to Farsio and Quade (2003), variations in unemployment likely lead to variations in output in the opposite direction. The authors argued that, although numerous factors cause variations in output, unemployment is the most visible because it has a direct impact on production. To a greater extent, many scholars believe that a reduction in the unemployment rate should increase the production of goods and services.
Two approaches were employed to estimate Okun's coefficient. The first is the output-gap approach, and the second involves using output growth and the first difference measure of the unemployment rate. Okun used the output-gap approach, where unemployment (UNEM) is dependent on deviations of actual output (GDP) from potential output (GDP). The relationship is specified as follows.
Assuming that the output gap is computed as the firstdifference of output (economic growth rate), then a new equation is obtained as follows: Conversely, Okun's law can be analyzed from the demand-side. Therefore, using the first-difference approach, output's response to variations in unemployment can be explained. The relationship can be expressed as follows: Equation 4

Unit root tests
Although economic theory requires a series to be stationary prior to estimating the relationship between the series, the ARDL approach used in this study does not require testing unit root for variables. Neverthelsss, it is important to perform the unit root test because the existence of a second order integration (i.e., I(2)) of any series ( The first portion of the above calculation includes changes (Δs) in the unemployment rate, the GDP growth rate and oil prices, and the second portion includes the GDP growth rate, the unemployment rate and oil prices.

ARDL bounds testing for cointegration
The ARDL method is utilized because of its advantages over other cointegration methods such as the residual-based technique (Engle & Granger, 1987) and the maximum likelihood test (Johansen, 1988(Johansen, , 1991Johansen & Juselius, 1990) we reject the null hypothesis that no cointegration exists between the series. If the F-statistic is less than the lower bound (I(0)), we accept the null hypothesis that there is no cointegration between the series. Furthermore, if the F-statistic falls between I(0) and I(1), our inference would be inconclusive.

Results and discussion
The results from bounds testing to cointegration are reported in Table 2. The results demonstrate that the computed F-statistic (11.129) is greater than I(1) at the 1% level. This result implies that a cointegrating relationship exists between the unemployment rate and economic growth (and oil prices).

Results of selected long run and short run models
Once the long run relationship between the variables had been confirmed, the ARDL model was estimated.
The optimal lag-length suggested by the Akaike information criterion (AIC) is (1,1,1). The long run and short run results for the selected models are presented in The coefficient of the error correction term lagged by one period (ECT t-1 ) is negative and statistically significant, and, therefore meets our expectation. The sign of the coefficient indicates a high speed of adjustment to equilibrium after a shock and indicates that approximately 92% of the deviations or disequilibrium in economic growth will be corrected within one year.

Results of diagnostic tests
The results reported in Table 4 Table 3. Results of the ARDL Model  Note: Δ is the first difference operator. Source: Author's computation using Eviews version 9.  Narayan (2005) (Table 5) suggest that the variables are cointegrated, because the computed F-statistic (5.338) is larger than I(1) at the 5% level.
The long run and short run results of the estimated ARDL model for 1972-2004 period are reported in Table 6. The optimal lag-length indicated by the AIC is (1,1,1). The long run results (Panel A) demonstrate that the unemployment rate has a significant and positive effect on the economic growth at the 10% level, but that oil prices do not have a significant effect on economic growth. The short run results (Panel B) reveal that unemployment does not have a significant effect on economic growth, but oil prices have a significant and positive effect on economic growth at the 5% level. The coefficient of the error correction term lagged by one period (ECT t-1 ) is negative and statistically significant, and therefore, meets our expectation. The sign of the coefficient indicates a high speed of adjustment to equilibrium after a shock and indicates that approximately 85% of the deviations or disequilibrium in economic growth will be corrected within one year.

Results of diagnostic tests
The results reported in Table 7   . The positive relationship between unemployment and growth reflects the situation in Nigeria, where individuals who are considered unemployed in the official sense still engage in certain types of employment that are not officially regarded as employment (Osinubi, 2005), and thereby contribute to production activity.

Conclusion
This study utilizes the ARDL bounds testing technique to examine Okun's law in Nigeria from 1970 to 2014, and considers the role of oil prices in the Nigerian economy. The results indicate a cointegrating relationship exists between the unemployment rate, economic growth, and oil prices. The results reveal that the unemployment rate has a long term negative and significant effect on economic growth, and that oil prices have a significant and positive effect on economic growth in the long term. The relatively low value of the coefficient suggests that Okun's coefficient is not only unstable, but also varies for different countries, and does not hold strictly for the Nigerian economy. The coefficient indicates that a 1 percentage increase in unemployment reduces economic growth by 0.18% in the long term, and vice versa. The long term negative effect of unemployment on growth has significant economic implications. An increase in the unemployment rate implies that more individuals are jobless and incomes are falling, which leads to decreased demand for goods and services and subsequently declining production.
All of these economic issues re-enforce unemployment problems over the long term. Therefore, policymakers should take steps to reduce unemployment to promote economic growth in Nigeria.